Skip to contents

R-package for Dynamic Path Analysis of survival data.

Overview

The dpasurv package is designed to perform dynamic path analysis via calls to the following principal functions:

  • dpa() fits the dynamic path models corresponding to a given dynamic path diagram; returns an object of class dpa
  • effect() estimates the cumulative direct and indirect effects from the fitted dpa-object; returns an object of class effect

The effect-objects can be summed together to obtain the total effect (i.e. direct effect + indirect effect) as well as plotted with bootstrap confidence intervals.

Installation

# To install the most recent stable version on CRAN:
install.packages("dpasurv")

# To get bug fixes or new features, install the development version:
devtools::install_github("Novartis/dpasurv")

Usage

A vignette on how to use the dpasurv package can be found here, and basic usage is summarized below:

library(dpasurv)

# Perform dynamic path analysis
s <- dpa(Surv(start, stop, event) ~ M + x, list(M ~ x), id = "subject", data = simdata, boot.n = 500)

# Extract direct, indirect and total effect
direct <- effect(x ~ outcome, s, alpha=0.05)
indirect <- effect(x ~ M ~ outcome, s, alpha=0.05)
total <- sum(direct, indirect)

# Plot the results
par(mfrow=c(1,3))
plot(direct); abline(h=0, lty=2, col=2)
plot(indirect); abline(h=0, lty=2, col=2)
plot(total); abline(h=0, lty=2, col=2)

# Plot the results with ggplot2 graphics:
ggplot.effect(list(direct, indirect, total))

Citation

The dpasurv package was created as supplementary code for the following manuscript:

@article{dpasurv,
  title={Dynamic path analysis for exploring treatment effect mediation processes in clinical trials with time-to-event endpoints},
  author={Kormaksson, M. and Lange, M. R. and Demanse, D. and Strohmaier, S. and Duan, J. and Xie, Q. and Carbini, M. and Bossen, C. and Guettner, A. and Maniero, A.},
  journal={Statistics in Medicine (Accepted)},
  volume={},
  number={},
  pages={},
  year={2024},
}

If you publish results obtained from the dpasurv package, please cite the above paper.

Code authors